Indicator targeting: Relies on non-income indicators that are meant to be correlated with poverty. These can include lack of or size of ownership of land, form of dwelling, social status, gender of head of household, etc.
Asymmetric information between the government, seeking to provide transfers to the poor, and individuals or households in the economy who can legitimately or otherwise seek these transfers, is the raison de ιtre of targeting. The underlying rationale of these targeting mechanisms is that administrative and other costs of identifying those who are poor are high, potentially reducing the resources that would be transferred to the poor under the scheme. Targeting mechanisms are a contractual/program-design innovation in response to the information asymmetry and the high costs of overcoming the information barrier.
However, this framework is implicitly less than comprehensive in approach, in the sense of focusing only on one scheme at a time. In a context where the principal (in a principal-agent
context) has several schemes in operation, the administrative costs per scheme (of overcoming information asymmetry) can get diluted substantially, thereby vitiating the need for indirect
targeting mechanisms for any specific scheme. Put alternatively, the issue of whether or not the administrative costs of identifying the poor are undertaken by the government usually does not
depend on any specific scheme. In an inter-temporal context, where the government does not know what specific schemes it may want to implement in near future, "tagging the poor"
Administrative Identification may provide externalities in terms of greater choices of schemes and their designs.
This is an important issue, as shown by the Indian experience where a large number of government poverty-targeted schemes rely on "Administrative Identification" (AI) to select beneficiaries. As shown in the table in Appendix 3, the most common criterion used in government schemes is that beneficiaries should be households below the poverty line (or BPL households). Other criteria, such as focusing on SCs/STs (which per se would represent
indicator targeting in the Indian context) are overlaid on the BPL status. As mentioned above, it may be argued that with an aggregate annual budget on CSS schemes exceeding Rs. 250
billion, it may be worthwhile for the government to undertake AI to better target the poor. Indeed, analytically it is perhaps more pertinent to ask why other targeting mechanisms should exist at all once AI has been undertaken. For example, some schemes listed in Appendix 3 rely on self-selection (e.g., food-for-work and rural employment scheme), geographical location, social category (SC/ST). Use of indirect targeting mechanisms in conjunction with AI may reflect in part the recognition that implementation of AI may be imperfect due to various reasons. In particular, the process itself may suffer from high Type I and Type II errors, as discussed below, resulting in exclusion of many poor and inclusion of many non-poor. In addition, the frequency of identification is necessarily spread apart in time, which would make it impossible to differentiate between transient and chronic poverty, (e.g., to differentiate the needy seeking food for work in face of natural calamity).
Administrative Identification: Tagging BPL (Below Poverty Line) Families
Since most PTPs currently in existence directly or indirectly rely on administrative classification of households into BPL and APL (Above Poverty Line), it is useful to briefly explain how this identification is undertaken. The exercise is intimately related to government efforts to provide food security to the population through the Public Distribution System (PDS). The PDS is a major component of aggregate subsidies spent by the GoI and is discussed more in Appendix 2.
The PDS, in its earlier forms, dates back to almost fifty years ago and was a general entitlement scheme with universal coverage until 1992. It provided rationed quantity of basic food (rice, wheat, sugar, edible oils) and some essential non-food items (kerosene oil and coal) at prices substantially below market prices. The central government was responsible for procuring, storing and transporting the PDS commodities up to central warehouses in each state/union territory, while the state government was responsible for distribution within the state.
While the universal coverage of PDS continued, the government introduced two major changes, the first in 1992, in the form of the Revamped PDS (RPDS) and, subsequently, in 1997 as the Targeted PDS (TPDS), both innovations targeted at poor households. The RPDS relied on geographical targeting, being introduced with universal coverage in only 1775 blocks in poor areas mainly tribal and hilly, drought prone and remotely located areas. The TPDS, on the other hand, was implemented in all areas but was open only to those identified as BPL. Along with the introduction of the TPDS, the price differential between PDS shops and open market was almost eliminated, effectively providing subsidized food only to BPL families.
At the core of the TPDS was division of the entire population into BPL and APL categories, based on the poverty line defined by the Planning Commission of India for different states for 1993-94. Multiple criteria were adopted for classification of BPL households, which in addition to income also included qualitative parameters like household occupation, housing conditions, number of earners, land operated or owned, live-stock, and ownership of durables such as TV, refrigerator, motor cycle/scooter, three wheelers, tractors, power tillers, combined threshers, etc. The responsibility for undertaking surveys and identifying the poor was with the state governments. However, the total number of BPL families in each state was capped somewhat arbitrarily at state-level estimates of the poor made by the Planning Commission using data for 1993-94, adjusted for growth in population in the interim.
Identified shortcomings of the BPL/APL targeting
Despite introduction in 1997, surveys for identification of BPL families were not completed in 18 out of 31 states by 2000 (CAG (2000)). Even in states where identification was completed, identification cards were not provided to a significant number of BPL families. Thus, implementation of the AI exercise has been slow and inefficient.12
A major criticism of the targeting is also that it has wrongly excluded a large number of eligible families. There are several reasons for this, both conceptual and operational. Conceptually, the main issue has been the appropriateness of income poverty to define the poor, specifically the absolute poverty line used by the Planning Commission. It is argued the official poverty line represents too low a level of absolute expenditure, which may exclude large sections of the population who experience low and variable incomes. If other criteria are used, such as nutrition, the number of households that can be deemed poor is much higher than
ceiling figures estimated by Planning Commission in 1993-94, (GoI (2002)).13
Operationally, as noted, identification surveys have not been completed in 18 of 31 states and, across the nation, 18 percent of families identified as BPL do not have identification cards. Even where surveys have been conducted, there still remain concerns on accuracy given the difficulties of measuring income. Since there are no regular official estimates of actual household incomes, implementation of BPL identification is subject to substantial practical and administrative problems. For example, an evaluation of the TPDS in Uttar Pradesh one of the poorest states in India by the World Bank based on the UP-Bihar Survey of Living Conditions (1997-98) found that 56 percent of households in the lowest income quintile did not get BPL cards. In the next quintile, 63 percent of the households were without the identification cards.
Thus, the AI exercise to classify all households into BPL/APL has been implemented with several shortcomings. Its progress has been slow, inefficient/corrupt and the results are not always reliable, with substantial errors of both type II and I. However, this exercise is used by a majority of the schemes in operation today that are targeting the poor households.
Download this Discussion Paper [ PDF 531.2KB| 53 pages ].
Post a Comment
We welcome your feedback on this publication. Post a comment. ADBI is not obliged to acknowledge or publish comments and may abridge or edit them before web posting.
There are  comment(s) for this entry. Post a comment.
The views expressed in this paper are the views of the authors and do not necessarily reflect the views or policies of the Asian Development Bank Institute (ADBI), the Asian Development Bank (ADB), its Board of Directors, or the governments they represent. ADBI does not guarantee the accuracy of the data included in this paper and accepts no responsibility for any consequences of their use. Terminology used may not necessarily be consistent with ADB official terms.